Power calculation for sensitivity of a diagnostic test

[Megan Moreton]

Hi,

I want to test the sensitivity of a rapid test & need to calculate the desired sample size. The sensitivity will be evaluated in a known population of positive individuals.

Could I calculate this using the following:

power oneprop 0.7 0.8, power(.8) onesided

This would be interpreted as: The study was powered to a sensitivity of 80% assuming a tolerable sensitivity of 70%. A total of 119 samples will be required to provide 80% power and a one-sided significance level of 0.05 to detect the sensitivity.

Thank you,
Megan

[Maria Sundaram]

Hello,

I am posting here because I would also find an answer to this question helpful. Megan, I don't suppose anyone replied to your answer in a message?

Thank you,
Maria

[Rich Goldstein]

I don't remember ever seeing this before but I doubt that what is in #1 is what is really wanted; maybe what is wanted is a narrow CI (but how narrow?) or maybe a test against some threshold value (but what is that value?)' maybe others can help based on what has been posted but I would want more information starting with what the exact null hypothesis is

[Megan Moreton]

Hi Maria - I can't remember exactly what I was looking at here, but I think in the end I used R to estimate the required sample size. I used the power.diagnostic.test function: https://www.rdocumentation.org/packa...iagnostic.test

I hope that helps!

[Maria Sundaram]

Thanks, Megan (and sorry for the late reply!)